210 ALASKAN GLACIER STUDIES
deposit from glacial streams emerging from the Variegated. In 1905 ice still existed beneath the flat, and alluvial deposit upon it was rapid. The advance of the Variegated Glacier in 1906 overrode a small part of the flat and destroyed one of the streams that had flowed out upon it in 1905. In 1909 the condition of the flat was not greatly different from the 1906 condition.
The interior flat seems to be a fairly constant feature of piedmont bulb glaciers, being developed just outside the mountain front. In the Butler Glacier, for example, the interior flat is just within a crescentic area of moraine-covered ice now completely detached from the glacier, and representing the last stages in the destruction of a piedmont bulb. Here no ice seems to exist beneath the former interior flat.
Galiano Glacier also has an interior flat area. The valley glacier expands beyond the mountain front and appears to terminate in a moraine-covered ice cliff; but beyond this is an area of lowland, which in 1890-91 was the seat of extensive, alluvial-fan deposit. The advance of Galiano Glacier prior to 1905 destroyed this flat by disturbing the ice beneath it, and also extended beyond the flat, raising the moraine-covered ice in a series of hummocks over a broad area. Evidently Galiano Glacier has much the same condition as Variegated Glacier except that the buried ice is thicker, and the alluvial fan on the interior flat was larger.
A perfect interior flat exists in the piedmont bulb of Allen Glacier in the Copper River valley, but here it is less advanced than in the cases already described, for alluvial deposit upon it has only just begun, and the ablation moraine on the inner side is less notably developed. It is possible that the lake in front of the Miles Glacier is developed on the site of an interior flat in that glacier. The Heney Glacier on Copper River also has an interior flat.
There is no well developed interior flat on the Lucia Glacier, though in 1905 and 1906 there was a small area of clear ice with a lower surface, in the midst of the ablation moraine, and just where the glacier began to expand notably. It was half or three quarters to a mile long and a quarter of a mile broad in its widest part, and offered the easiest route over the otherwise moraine-covered, hummocky surface of the glacier between Terrace Point and Floral Pass. This was being destroyed by the 1909 advance.
In 1905 and 1906 there was no interior flat area on Atrevida Glacier, but a very perfect, well-defined area of clear ice was introduced by the 1906 advance and, by the progress of ablation in the interval between its formation and our next visit in 1909, its clear surface was lowered well below the level of the surrounding area of moraine-covered ice. As a result of the 1906 advance a similar area, though not as well defined because not so thoroughly moraine enclosed, developed on Variegated Glacier some distance inside the previously-existing flat.
The interior flat is so frequent an associate of the piedmont bulb condition that we believe it to be due to the operation of some general cause. The nature of this cause is suggested by the development of the area of clear ice in Atrevida Glacier during the 1906 advance. As stated in the discussion of this phenomenon there is strong reason for considering it the result of the upflow of clear ice from below, when, during advance, the flowage of the ice is retarded by the resistance of the stagnant, partly rigid, outer portion of the piedmont bulb.
For the phenomenon we propose the following working hypothesis, whose fuller discussion may be postponed until we have a larger body of observational data from a